Background: ‘Atypical’ antipsychotic agents (AAPs) are used to treat the symptoms of schizophrenia. Seven AAPs are currently on the market in the United States and Canada: aripiprazole, clozapine, olanzapine, paliperidone, quetiapine, risperidone, and ziprasidone. The current literature provides head-to-head comparisons for only 13 of the possible 21 pairwise comparisons. A complete summary of the evidence would require indirect comparisons, which can be imprecise and cumbersome to implement. A more unified approach is to simultaneously model all comparisons using a mixed treatment comparisons (MTC) model. In addition, studies of the same drug utilized a range of dose levels making simple comparisons dubious. We required a framework that could take into account dose levels and other study-level covariates. Objectives: We investigated the practicality of using the MTC model in a meta-regression framework to compare the seven AAPs among adults with schizophrenia on the outcomes of all-cause discontinuation and discontinuation due to adverse events. Methods: We searched the Cochrane Central Register of Controlled Trials, MEDLINE, and PsychINFO through November 2007 for relevant head-to-head trials of at least two AAPs. An MTC meta-regression model was estimated under a Bayesian framework. Possible covariates included drug dose level of each treatment arm, study quality, study sponsorship, patient population, and study group disease severity based on the Clinical Global Impression Scale. Results: Our search identified 67 relevant head-to-head trials, including multi-arm trials. The largest number of direct comparisons was between olanzapine and risperidone. Preliminary analyses controlling for dose level indicated that clozapine and olanzapine had lower drug discontinuation rates than aripiprazole, olanzapine, quetiapine, risperidone and ziprasidone. We also found higher rates of discontinuation due to adverse events with clozapine compared to olanzapine, quetiapine, or risperidone. Further work with this model continues. Conclusions: The most comprehensive trial of AAPs included olanzapine, quetiapine, risperidone and ziprasidone. The MTC model can provide a unified summary of the comparative effectiveness of all drugs in the class. Our case study shows that it is feasible to extend the model to adjust for treatment arm-level and study-level covariates.